Time-resolved absorption and fluorescence spectroscopy are used to study the dynamics of protein structural changes subsequent to rapid mixing or excitation with short laser pulses. Kinetic models are used to fit and interpret the measured data. We have focused our recent efforts on understanding the folding of the villin subdomain in more detail. This 35-residue domainconsists of three helices which interact to form a hydrophobic core. To probe the folding mechanism we have carried out structural and physical chemical studies of molecules with point mutations and other sequence modifications. In one series of experiments we carried out 'conservative' mutations which only perturb residues by addition of deletion of a methyl group. In the villin sequence these are limited to L1V, A16G, L20V, Q25N and L28V. In addition we have studied the effects of deleting the charges of the lysine residues at positions 24 and 29 by replacing the amino group with a methyl group (norleucine). These studies have included determination of the structures of some of the folded states by x-ray diffraction carried out in collaboration with Dr. Thang Chiu and David Davies in the LMB/NIDDK. The high resolution x-ray structures of the wild type protein show that the L24 and L29 residues are mostly buried, passing through the hydrophobic core of the folded protein with the charges emerging on the protein surface. The structures of the modified sequences exhibit only local rearrangements of the hydrophobic side chains which interact with K24 and K29. Dr. Chiu has also crystallized a fragment of the villin subdomain consisting of residues 1-20, which contains the first tywo helices and all of the core Phe residues. In the presence of small amounts of trifluoroethanol, this fragment crystallizes in a complex crystal in which 50% of the molecules having the native fold. None of the conservative mutations have significant effects on the folding rates. Similar results had already been observed for substitutions of a number of other residues, including the three core phenylalanine residues (F6L, F10L and F18L), the C-terminal phenylalanine, F35A and alanine 18 (A18S, A18V). Together, these results suggest that the transition state for folding of villin is close to the denatured ensemble of conformations. On the other hand, the K24Nle and K29Nle mutations both stabilize the folded structure and accelerate folding by factors of ~2.5, showing that both mutations stabilize the transition state on the folding pathway. The double mutant, with a folding time of ~ 700 ns, ranks as the fastest folding protein studied to this time. Since both of these mutations increase the hydrophobicity of the third helix, one possible explanation of the results is that the formation of a turn in the region of residues 20-24 is rate-limiting. Hydrophobic interactions between the hydrophobic moieties on helix 3, including the Nle residues, with the hydrophobic residues on helices 1 and 2, iincluding F5, F10 and F17, favors the formation of this turn.